ABSTRACT The endosomal system internalizes membrane and protein from the plasma membrane, and then in a topologically distinct process internalizes membrane into itself to form multivesicular bodies (MVBs). Internalization of membrane and associated cargo into the MVB is essential for digestion and degradation in the lysosome, and is also used by specialized cells to generate exosomes. A set of eighteen "class E" proteins conserved from yeast to mammals is thought to regulate and probably drive the formation of intralumenal vesicles. Some or possibly all of these proteins are also used by nonlytic viruses such as HIV to bud from the plasma membrane of infected cells in a process that is topologically equivalent to budding into the lumen of the endosome. Elegant studies in yeast and more recently mammalian systems have lent insight into how these proteins associate with each other, cargo molecules, and the endosomal membrane, but essentially nothing is known about how these or other proteins drive membrane invagination and vesicle release. Because the required membrane curvature is opposite to that of traditional coated vesicle-driven endocytosis, it is likely that novel mechanisms are involved. We found using quick-freeze deep-etch electron microscopy that ESCRT-III proteins related to Snf7 (hSnf7/CHMP4 in mammalian systems) assemble into filaments on the plasma membrane that can be induced to form tight circular arrays and bend the membrane away from the cytoplasm, creating buds and eventually tubules on the cell surface. We hypothesize that similar ESCRT-III containing polymers normally create the neck of nascent endosomal vesicles, both confining the vesicle's contents and inducing the requisite deformation in the membrane. Our plan is to define the structure and dynamics of ESCRT-III polymers in vitro, on model membranes, and in the context of normal MVB biogenesis in order to determine how assembly and disassembly of ESCRT-III polymers participate in MVB biogenesis. Aim 1 will explore the structure of ESCRT-III homo- and hetero- polymers. Aim 2 will define mechanisms that regulate ESCRT-III polymer assembly. Aim 3 will study disassembly of ESCRT-III polymers by the AAA+ ATPase VPS4. Finally, Aim 4 will study endogenous ESCRT-III proteins in cells during the process of receptor downregulation before and after silencing expression of VPS4 and hSnf7 proteins.